JOURNAL OF APPLIED TOXICOLOGY, VOL. 12(6), 407414 (1992)
Magnetic Resonance Imaging of the Rat Brain Following Acute Carbon Monoxide Poisoning Vishram Jalukar,' David G. Penney,'*t Michael Crowley2 and Nicholas Simpson2 'Department of Physiology, Wayne State University School of Medicine, Detroit, MI 48201, USA 2Vaitkevicius Magnetic Resonance Center, Harper Hospital, Detroit, MI 48201, USA
Key words: blood pressure; brain; carbon monoxide; cerebral; edema; glucose; hematocrit; hypotension; hypothermia; lactate; magnetic resonance imaging.
Magnetic resonance (MR) may be used for repeatedly and non-invasively imaging the brain. Until now, no studies have used this approach to study the effects of carbon monoxide (CO) poisoning in a defined animal model. Conscious, Levine-prepared female rats (unilateral carotid artery and jugular vein occlusion) were exposed to 2400 ppm CO for 90 min, with or without the infusion of 50% glucose solution; CO-stimulated increases in blood glucose and lactate occurred in both groups, while blood pressure and body temperature fell. One to four hours following termination of CO exposure, increased cortical pixel intensity, cortical surface area and brain midline shift were observed on the operated side of the brain in some rats of both groups (i.e. responders = R), providing evidence of edema. At sacrifice, 5 h following termination of CO exposure, gross water content was increased on the left side in the corresponding cortical slices in R rats, providing another measure of edema. Significant positive correlations were found between left to right pixel intensity difference and water content difference, and between the extent of midline shift and water content difference. The elevations of blood glucose and lactate concentrations, and the magnitudes of CO-induced hypothermia and hypotension were similar to those in past studies, but appeared to exert no effect on the severity of cortical edema in terms of differences in pixel intensity, surface area, midline shift or gross tissue water content. Thus, the observed differences between the R rats and the non-R rats is not explained by the available data. The results of this study demonstrate that MR imaging can detect changes in the cerebral cortices of rats given an acute toxic challenge with CO. Moreover, in responders the edema develops quickly, reaching nearly full development within 1 h after CO exposure.
INTRODUCTION
Carbon monoxide (CO) is a poisonous gas that causes death, as well as a wide variety of immediate and delayed morbid effects, with nervous system damage being the most common. In the rat, acute CO poisoning is characterized by hypotension, bradycardia, hypothermia, hemoconcentration and hyperglycemia. 1-3 A recent study using the Levine-prepared rat demonstrated increased neurological deficit and general morbidity following 90 min of C O exposure in subjects experiencing elevated blood g l u c o ~ eSimilar .~ findings were obtained from a retrospective study of human CO poisonings.s Thus, glucose level appears to be an important determinant of cerebral viability and overall survival during the hypoxia and the relative ischemia of C O poisoning. Whether elevated blood glucose permits an increased rate of glycolysis, generating more lactate and hence an increased blood and tissue acidosis, is unknown. Magnetic resonance (MR) is a recent, non-invasive technique for assessing brain structure, chemistry and the damaging effects of hypoxia, ischemia and toxic agents. Through the use of spectroscopic applications of nuclear MR, brain levels of high-energy phosphates
t Author
to whom correspondence should be addressed.
02~37W9U060407~8$09.00
01992 by John Wiley & Sons, Ltd.
have been assessed in the rat and other animal Recently, MR imaging techniques were used to study structures as small as the rat heart, following chronic C O exposure.8 While MR has been used to assess brain damage following human C O poisonings,*-" it has not been applied previously to study experimentally the induction of central nervous system damage in an animal model of CO poisoning. Some reports suggest that the technique may be superior to computerized tomography (CT), by the earlier detection of cerebral ischemic changes. l2 This study was designed to use MR imaging for non-invasively and repeatedly providing detailed information on the cerebral cortices of acutely COpoisoned rats a short period of time following termination of CO exposure. The study was carried out in an animal model in which the metabolic, cardiovascular and neurological responses to acute severe C O poisoning have been characterized previously. A major goal was to investigate the reputed deleterious role of elevated blood glucose resulting from CO exposure. A second goal was to determine whether changes in blood lactate, body temperature, blood pressure and hematocrit are associated with development of brain edema, as imaged with MR techniques. A third goal was to determine whether the magnitude of cortical edema assessed by MR methodology correlates with measurements of brain gross water content. 143'3,14
Received 12 December I991 Accepted (revised) 17 March 1992
408
V . JALUKAR ET A L .
MATERIALS AND METHODS
Table 1. Characteristics and settings of the magnetic resonance imaging instruments
Animal acquisition and treatment Female Sprague-Dawley rats, 90-120 days of age, were used. Two days prior to C O treatment, a modified Levine p r e p a r a t i ~ n ' . 'was ~ produced. The rats were anesthetized by intramuscular injection of ketamine (0.8 ml kg-')-Rompun (0.24 ml kg-'). The left common carotid artery and jugular vein were catheterized with PE-50 polyethylene tubing, placing the left cerebral hemisphere at greater ischemic risk. Both catheters were threaded subcutaneously to the nape of the neck, tied in place and plugged with Amphenol pins. Catheters were flushed twice daily with 0.1 ml of a heparin-saline solution (50 U heparin ml-*) to maintain patency. The rats were allowed to recover under close observation: they were not behaviorally different from unoperated controls. Rats ( n = 6) were exposed to 2400 ppm C O for 90 min in the conscious state. They were loosely confined in a transparent acrylic restrainer, inside a large transparent plastic bag located in a fume hood. The CO concentration was determined continuously by a Beckman model 870 infrared C O analyzer. Rectal temperature was monitored by a YSI Tele-thermometer model 43TD, using YSI 400 series probes. Heart rate and blood pressure were monitored at 0,45 and 90 min of CO exposure by a Gould model 2400 ink pen recorder, with Statham P23id pressure transducers connected to the carotid catheter. Small blood samples were withdrawn from the carotid catheter at the above times for measurements of glucose, lactate and hematocrit. A YSI 2300 STAT glucosek-lactate analyzer was used to determine blood glucose and lactate concentrations. The hematocrit was determined in duplicate by the microhematocrit method. Hyperglycemia was induced in rats ( n = 6) by the infusion of exogenous glucose (50% w/v in 0.9% saline) at a dose of 6.0 ml kg-' body weight, as described earlier.' The glucose was administered through the jugular catheter using a Harvard Apparatus Co. infusion/withdrawal pump. Infusion rates were adjusted to deliver 67% of the glucose solution in the 15 min preceding CO exposure, and the remaining 33% during CO exposure. Control rats (n = 3) were modified surgically but not treated with glucose and/or CO. Magnetic resonance imaging Immediately following CO exposure, the animals were lightly anesthetized with ketamine (0.5 ml kg-')Rompun (0.16 ml kg-') and taken to the Magnetic Resonance Center of Harper Hospital. A few unoperated control rats were also imaged under light anesthesia. The study was initiated using the 1.5 Tesla Siemens GBSII MR imaging instrument. Owing to hardware and software failures requiring protracted repairs, it was decided to complete the study using the 4.7 Tesla Bruker Biospec MR imaging unit. Rat brains were imaged in both instruments using Alderman-Grant slotted resonator coils. The instruments were autoadjusted for frequency, receiver gains and transmission power (transmission power was atten-
Parameters
Siemens GBSll 1.5
Bruker Biospec 4.7
Imaging parameters Field strength ( T ) Coil diameter (cm) Coil length (cm) Tuned (MHz) Field of view (cm) In-plane resolution (mm) Slice thickness (mm)
1.5 10.2 10.2 63 8.9 0.349 3.0
4.7 8.2 17.3 200 9.0 0.352 3.0
Spin echo imaging TE1 (ms) TE2 (ms) Slice distance (mm) Slice gap (mm) Acquisitions Imaging time (min)
22 or 26 90 0.05 0.15 2 21
26 90 1.75 5.25 1 17
uated 25 dB subsequent to autoadjust in the Bruker instrument) (see Table 1). Scout images were used to set the orientation of the slices in the spin echo sequence in order to obtain true coronal images, as shown below. Spin echo images were acquired 1, 2, 3 and 4 h after removal from CO. Three adjacent images corresponding to rostral, middle and caudal sections of the brain were analyzed for midline shift, total cortex area and relative edema index, using standard Siemens or Bruker software. Midline shift was calculated by measuring the diameter of the total brain and of each hemisphere. The relative edema index was calculated by measuring the average pixel intensity within a circle of radius 0.210 cm from corresponding portions of each cortical hemisphere. Total cortex area was defined by drawing a freehand outline of each cortical hemisphere. Rats were sacrificed after imaging by administering an overdose of ketamine through the jugular catheter. The cortex was immediately removed by dissection and sectioned into three slices per hemisphere. The sections were dried at 60°C for 48 h. Wet and dry weights were measured using a Mettler H35AR analytical balance. Data analysis was carried out on Macintosh microcomputers using Excel and Cricket-Graph programs. Most values are means f SEM. Student's #-test was used for statistical analysis, with P values =s 0.05 considered to be significant. RESULTS Upon 'H imaging, some CO-exposed rats of both groups (CO alone and CO+glucose) showed marked edema and swelling of the left cerebral hemisphere ('responders'), while others did not ('non-responders'). The presence of cerebral edema is indicated by increasing brightness in the region of fluid accumulation and of a rightward shift of the brain midline (Fig. 1C,
MR IMAGING IN CO POISONING
E). This is illustrated by the plots of pixel intensity superimposed upon each cortical cross-section (Fig. l A , B, D, E). Control rats, non-responders and unoperated animals (not shown) showed no right versus left brightness or midline differences (Fig. l A , B, D). The photographs shown were judged to be representative of the responses by controls, responders and nonresponders. Note that, by convention, the left side of the brain is on the right side in each photograph. The data presented are a fraction of the total number of rats treated and imaged. Some rats were not imaged satisfactorily and thus were not included in the final data set. No selection criteria were applied other than completeness. Evidence of cortical edema is also seen in Fig. 2 for individual animals, in terms of intensity (A), midline shift (B), differences in area (C) and in water content (D), which was determined afterwards on corresponding brain slices. The values for intensity, midline shift and area at 1 and 4 h of recovery, and for water content ( 5 h only). are seen in Table 2. There was no clear time dependence with regard to the development of cerebral edema in responders, either in terms of the differences in intensity or in midline position (Fig. 3A, B). In most cases, increased intensity and the midline shift were already present in responders by 1 h of recovery. Correlations of both intensity difference and the midline shift at 1 h of recovery with the water content difference of the corresponding brain slices were each statistically significant (Fig. 4A, B). The correlation of midline shift at 4 h with water content difference was also significant. When considered in terms of mean values, the responders tended to show area and intensity difference values greater than the nonresponders (but not significant) (Fig. 5). The water content difference of the responders was significantly elevated above the non-responders (Fig. 6). There was no difference in water content between the nonresponders and a large group of unoperated controls. The metabolic and physiological parameters of the experimental subjects are shown in Table 3. They are displayed in two ways: as rats that received CO alone and CO+glucose, and as responders and nonresponders. Glucose infusion raised blood glucose to 241 mg d1-I 45 min after beginning CO exposure and to 168 mg dl-' in rats receiving CO alone. Blood lactate in the rats receiving glucose was raised to 170 mg dl-' 45 min after beginning C O exposure and to 151 mg dl-' in rats receiving CO alone. Average blodd glucose and lactate values of all seven animals prior to CO exposure were 97 and 9 mg dl-', respectively. Also at this time, mean arterial blood pressure was 110 mmHg, body temperature was 37.6"C. heart rate was 470 beats min-' and the hematocrit was 41.7%. The responder group consisted of two CO and two CO+glucose rats, while the non-responders consisted of one CO and two CO+glucose rats. There were no significant differences ( P > 0.05) between the responders and the non-responders with respect to blood glucose or lactate concentrations, or change in mean arterial blood pressure, body temperature. heart rate or hematocrit (Table 3). None the less, lactate concentration and the fall in blood pressure and body temperature tended to be greater in the responders.
409
Neither blood glucose nor lactate concentration or hypotension during CO exposure appeared to be correlated with the severity of cortical edema. ~~
~~
DISCUSSION
The results of this study demonstrate that MR imaging can detect changes in the cerebral cortices of Levineprepared rats acutely poisoned with CO. The edema develops quickly, being well developed within 1 h after exposure ends. It develops in this model with or without the addition of exogenous glucose, and is apparently not related to the magnitude of hyperglycemia, lactic acidemia, hypothermia, hypotension or hemoconcentration. Thus, the observed differences between the responders and the non-responders is not explained by the available data. On the other hand, even if correlations had been established, it would not be possible to state with certainty that the changes in these parameters were responsible for the cortical edema, or that they resulted from it. It is possible that the non-responders had adequate bilateral cortical perfusion despite the surgical modification, thus obviating the development of unilateral cerebral hypoxia sufficient to result in edema. This may be attributable to the anatomical variations among rats. Explanations involving differences in oxygen delivery, metabolism, etc. are also possible. Significant associations of edema development with cardiovascular and other physiological parameters may resolve if a larger number of animals were to be studied. Hypothermia, hypotension, bradycardia, hemoconcentration and altered blood glucose concentration are the usual effects of acute severe CO poisoning in the The loss of normal body temperature results from decreased heat production through an inhibition of oxidative metabolism and an increased heat loss due to peripheral vasodilation. l 3 Decreased arterial blood pressure is a function of the peripheral vasodilation, and depressed pump function is due to both decreased heart rate and stroke volume. The rise in the hematocrit is presumed to be due to the loss of plasma volume resulting from an increased endothelial permeability ."j Increased hematocrit and lowered body temperature act to increase blood viscosity. This, along with compromised cardiac function, contributes to inadequate cerebral perfusion. In the Levine-prepared rat, the oligemia is more extreme during CO exposure on the operated side of the brain, leading to edema and behavioral evidence of unilateral brain dysfunction. I Previous studies in this laboratory showed that elevated blood glucose is associated with poor outcome in the CO-poisoned rat."' This was the case whether the elevated glucose resulted from CO exposure' or from the infusion of exogenous glucose.3 The blood glucose concentrations achieved during CO exposure in the present study in the C O alone and CO+glucose rats were of a magnitude similar to earlier studies. None the less, it should be noted that the responders showed no higher blood glucose concentration than the nonresponders. Elevated blood glucose is also reported to be deleterious in human acute CO p o i s ~ n i n g . ~ The mechanism by which hyperglycemia produces
410
V . JALUKAR ET A L
Figure 1. TZ-weighted proton magnetic resonance echo 1 images of rat cerebral cortices: (A) Control; (6) CO-exposed non-responder; (C) CO-exposed responder; (D) CO-exposed + glucose-treated non-responder; (E) CO-exposed + glucose-treated responder. All but control are 4 h post-CO exposure. The left side of the brain is on the right side of the picture. All but image C were acquired on the Bruker 4.7T instrument: image C was acquired on the Siemens 1.5T instrument. Pixel intensity is plotted in all but panel C.
41 1
MR IMAGING IN CO POISONING
I
100, 40
A.
30 20
80 A
0
a
-
s - .
10
m
730.1 807.2
0 812.1
60-
.-c
-10
40-
a
.
s
20-
w
a
I
4
-20
0
2
1
3
5
4
8
B. 61
I
-204
I
a o -4
1
0
co
Control
3
2
5
4
Recovery Time (hrs.)
CO + Glucose
Figure 3. Differences between left and right cerebral hemispheres of rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose, at 1,2,3 and 4 h post-exposure: (A) mean pixel intensity; (6)midline position. Responders are denoted by filled symbols and non-responders by open symbols. Numbers are animal identifiers.
Figure 2. Differences between left and right cerebral hemispheres of non-exposed control rats and of rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose: (A) mean pixel intensity; (6)midline position (C) total cortical area; (D)water content. (A-C) 1 h postC0 exposure; (D)ca. 5 h post-CO exposure.
Table 2. Differences between left and right cerebral hemispheres of control rats and rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose twice during recovery Control rats
CO rats
CO+glucose rats
Parametermime
702
A
721
724
730-1
807-1
807-2
812-1
-2.11
4.12
12.79 19.78
30.17
-10.09 5.36
41.38 86.36
2.13 7.61
10.17 37
1.4 9.33
1h 4h
0
0.68
3.95 5.32
1.52 3.44
0.71 0.34
3.62 5.47
1.57 1.27
2.53 3.95
-2.52 0
Area
I h
6.36
-16.12
(YO)
4h
35.49 40.96
5.2 14.86
0 -3.66
39.63 55.1
15.84 -9.46
12.09 13.24
-3.74 7.25
0.06
0
4.22
3.20
-0.49
4.45
0.43
3.07
0.31
A Midline (YO)
A
708
1h 4h
Intensity
l%)
A
821
Water contenta (YO)
Water content was determined once at ca. 5 h of recovery.
brain damage has been tied to the accumulation of lactate and to the resulting acidosis. l4 Acidosis could be more severe in CO poisoning than in pure ischemic situations, since some blood flow continues to supply glucose and to generate lactate, although the flowing blood also continues to wash the lactate away. In other models, incomplete ischemia has been shown to have a worse outcome than complete ischemia.17 In any event, blood lactate concentration was observed to increase some 16-fold during CO exposure in the present study,
from 9 mg dl-1 (pre-CO) to ca. 150 mg dl-* (during CO). This must have been accompanied by a sharp fall in blood pH; unfortunately, we did not measure pH. SokalIs found that blood lactate increased from 9 to 62 mg dl-' in rats exposed to 4000 ppm CO for 40 min, or an 18-fold increase. Considering that the CO, concentration that he used was nearly double that of the present study, the similarity in response was excellent. Blood pH fell to 6.85. Similar changes in blood lactate were reported by McGrath et ~ 1 . in ' ~rats
412
V. JALUKAR ET A L . I
1001
80
C 0,
1 .
0
5.0
..
r = 0.86. WO.01
-20 -0.01
0.00
0.01
0.02
0.03
0.04
s i -1.01
0.05
Responders
Nonresponders
Controls
Figure6. Cortical water content of rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose, and non-exposed controls. Exposed rats have been grouped as responders and non-responders. The numbers of rats in each group, and the means t SE are given. Comparing responders to non-responders and controls; ** P < 0.01.
-001
000
0 01
0 02
0 03
0 04
0 05
A Water Content (%I
Figure4. Regression plots using data from left and right cerebral hemispheres of rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose, and of nonexposed controls: (A) difference in mean pixel intensity (lef’t minus right) at 1 (open symbols) and 4 h (closed symbols) post-CO exposure vs. the difference in cortical water content (left minus right); (B)differnece in midline position vs. the difference in cortical water content (left minus right). Intensity and midline position data obtained at 1 (open symbols) and 4 h (closed symbols) post-CO exposure; water content obtained at ca. 5 h post-CO exposure. Midline position was calculated by subtracting the radius of the left hemisphere from the radius of the brain.
80 70
T
I
-30
Responders
Nonresponders
Controls
Figure5. Cortical area and mean pixel intensity of the cortex in rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose, and non-exposed controls, at 1 and 4 h post-exposure. Values for controls determined once. The CO-exposed rats are grouped as responders and nonresponders. The numbers of rats in each group are given. Vertical bars indicate standard error.
acutely inhaling CO. Using mice inhaling 3500 pprn CO, Moore et al.”’ found that blood lactate reached 90 mg dl-’ after 11.5 min while pH fell to only 7.25, the more modest acidosis probably being due to the short duration of exposure. Measurements of regional brain lactate concentration have shown it to increase to similar high levels in anesthetized rats ventilated with up to 20000 ppm CO.” In the past, a number of investigators have suggested that hypoxic/ischemic brain damage is the direct result of excessive lactate production and the attendant acidosis,22 and that brain lactate can be used to calculate brain pH. Recent MR studies of brain lactate and pH, however, suggest that the two become dissociated during hypoxia/ischemia.23.24Although there were no differences in the blood lactate levels in our rats, it is still possible that brain pH was different in the responders versus the non-responders. Nevertheless, recent studies in our laboratory have found no difference in peak blood lactate concentration between rats that died during and following CO exposure and those that survived, whether or not exogenous glucose was given (Sutariya L? Penney, unpublished data). Administration of 50% glucose solution, which raised the blood glucose level to 325-450 mg dl-’ during CO exposure, failed to elevate further the peak blood lactate level. This finding corroborates the present study, where the administration of glucose failed to elevate significantly the blood lactate level and where responders showed no higher lactate level than non-responders. Of course, it is possible that tissue lactate concentration in the cortical region that sustained edema departed significantly from that of the carotid blood. We found an increase in cortical water content of as much as 4.5% on the operated side of the brain in responder rats. This value agrees well with an earlier study with Levine-prepared rats exposed to 2700 ppm CO, in which the difference in cerebral water content (left minus right) was significantly correlated with behaviorally derived neurological scores. * The impaired rats showed obvious evidence of unilateral brain dysfunction, e.g. one-sided weakness and circling. Neurological assessment in a similar fashion was not possible in the present study because of the necessity to anesthetize the rats during
413
M R IMAGING IN CO POISONING
Table 3. Metabolic and physiological parameters of rats exposed to 2400 ppm carbon monoxide alone and 2400 ppm carbon monoxide + glucose, of the rats of both groups that showed evidence of cortical edema (responders) and of rats not showing cortical edema (non-responders)." Parameterrrime (min)
CO (3)
CO+Glucose (4)
Responders (4Ib
* 14.0
Non-responders (3)c
Glucose (mg dl '1 Lactate (mg dl-')
+45
167.7
33.2
241 t 50.6
180.8
+45 +90
151.0 2 12.7 137.0 ? 20.0
69.8 t 19.0 50.5 ? 13.3
158.3 15.7 154.0 5 13.1
66.3 ? 21.7 32.3 5 18.0
A
+90
-36.0 5 1.0
45.4 5 4.5
-43.5
4.4
39.8
9.5
- 1 4 t 33.8
Av. blood
pressure (mmHg) A Heart rate (beats min-'1
A A
+90
-43
?
?
17
Body temp. ("C)
+90
-4.2 t 0.3
Hematocrit
+90
5.1 f 0.5
2.3 2 99.3 -3.4 5 0.4 6.0
?
-19.5
-4.2 t 0.2
* 0.6
5.2
t_
0.3
248
-3.0
?
?
?
78.0
7.5
0.4
6.2 5 0.9
(YO)
a
Values are means t SEM. Responders: two CO and two CO+glucose. Non-responders: one CO and two CO+glucose.
MR imaging and immediately prior to cortical tissue removal for the measurement of water content. The finding in normal rats of constant cortical water content, with very little variability between the two hemispheres, suggests little influence of potentially confounding factors (e.g. estrous cycle). The brain is relatively resistant to injury due to hypoxia, but edema develops almost immediately in response to ischemia." Hypoxia is almost always accompanied by a decrease in blood pressure. In fact, severe hypoxia leads to cardiovascular failure resulting in global ischemia, thus the effects of hypoxia alone are difficult to assess.'5 In cases where blood pressure is maintained, there is no evidence of brain abnormalities in experimental animals subjected to hypoxia.2h Brain edema results from the retention of water.27 Studies on ischemia suggest that there is a threshold phenomenon for edema formation.28One might expect to find a correlation between the CO-induced hypotension observed and edema formation. The absence of such a correlation in our rats indicates that there may be variability between individuals in the collateral circulation. The existence of similar collateral circulatory variability was inferred in mice exposed to C O or ether.2' Further study of cerebral blood flow during CO exposure is indicated. We have shown that hypothermia. hypotension and hemoconcentration are usual events during CO exposure in the rat,' and indeed that hypothermia confers protection in terms of survival and neurological outcome."' Inexplicably, in the present study, body temperature fell by more in the responders than in the non-responders, although the difference was of marginal statistical significance. Two human cases of CO-induced brain damage reportedly have been diagnosed with MR imaging, where lesions were found in the globus pallidus: in one case, bilateral cortical damage also was found." In a second case report, damage to the globus pallidus caused by
CO exposure was determined using both MR and CT imaging techniques.'" More recently, Tuchman et al. using MR imaging, found bilateral increased signal intensity in the anterior thalamus of a child poisoned with CO. An earlier CT scan had revealed no abnormality, suggesting the advantage of MR imaging. Other studies" suggest that MR imaging may reveal cerebral ischemic changes earlier than other techniques. In summary, this study of acute severe C O poisoning in the Levine-prepared rat resulted in the following findings: (i)
development of edema as evidenced by increased pixel signal intensity, surface area and midline bulging of the cerebral cortex on the operated side of the brain, 1-4 h following the termination of CO exposure in responder rats; (ii) development of edema as evidenced by increased gross water content in the corresponding cortical slices in responder rats, approximately 5 h following the termination of CO exposure; (iii) demonstration of a significant positive correlation between cortical pixel signal intensity difference as a measure of edema and water content difference and between midline shift as a measure of edema and water content difference; (iv) within the limitations posed by the small number of animals used, the increase in blood glucose and lactate showed no correlation with the severity of cortical edema, as judged by differences in pixel signal intensity, surface area, midline position or gross water content.
Acknowledgements This work was supported by an American Heart Association of Michigan Summer Research Fellowship to Mr Vishram Jalukar. We wish to thank the Magnetic Resonance Center, Harper Hospital, for providing the instrument time to carry out this study. We also wish to thank William Negendank, M D and Jeffrey Evelhoch, PhD.
414
V. JALUKAR ET A L .
REFERENCES
1. D. G. Penney, K. Verma and J. A. Hull, Cardiovascular, metabolic and neurologic effects of acute carbon monoxide poisoning in the rat. Toxicol. Lett. 45, 207-213 (1989). 2. D. G. Penney, P. Sharma, 8. B. Sutariya, and 6. G. Nallamothu, Development of hypoglycemia is associated with death during carbon monoxide poisoning. J. Crit. Care 5, 1-7 (1990). 3. D. G. Penney, C. C. Helfman, J. C. Dunbar and L. E. McCoy, Acute severe carbon monoxide poisoning in the rat: effects of hyperglycemia and hypoglycemia on mortality, recovery and neurologic deficit. Can. J. Physiol. Pharmacol. 69, 1168-1177 (1991). 4. D. G. Penney, C. C. Helfman, J. A. Hull, J. C. Dunbar and K. Verma, Elevated blood glucose is associated with poor outcome in the carbon monoxide-poisoned rat. Toxicol. Lett 54, 287-298 (1990). 5. D. G. Penney, Hyperglycemia exacerbates brain damage in acute severe carbon monoxide poisoning. Med. Hypoths. 27, 241-244 (1988). 6. W. Auffermann, N. Derugin, S. Wu, W. W. Parmley, C. B. Higgins and J. Wikman-Coffelt, Effect of reduced perfusion pressure in cardiac energetics and function in pressure overload hypertrophy -3'P-NMR in rats. Circulation, Suppl. 76 (IV), IV-439 (1987). 7. K. L. Behar, J. A. den Hollander, 0. A. C. Petroff, H. P. Hetherington, J. W. Prichard and R. G. Shulman, Effect of hypoglycemic encephalopathy upon amino acids, highenergy phosphates, and pHi in the rat brain in vivo: detection by sequential 'H and 31P NMR spectroscopy. J. Neurochem. 44, 1045-1 055 (1985). 8. G. A. Bambach, D. G. Penney and N. G. Negendank, In situ assessment of the rat heart during chronic carbon monoxide exposure using magnetic resonance imaging. J. Appl. Toxicol. 11, 43-49 (1991). 9. A. L. Horowitz, R. Kaplan and G. Sarpel, Carbon monoxide toxicity: MR imaging in the brain. Radiology 162, 787-788 (1987). 10. G. Meucci, G. Rossi and M. Mazzoni, A case of transient choreoathetosis with amnesic syndrome after acute carbon monoxide poisoning. ltal. J. Neurol. Sci. 10, 513-517 (1989). 11. R. F. Tuchman, F. G. Moser and S. L. Moshe, Carbon monoxide poisoning: bilateral lesions in the thalamus on MR imaging of the brain. Pediatr. Radio/.20,478-479 (1990). 12. J. T. Sipponen, M. Kaste, R. E. Sepponen, T. Kuurne, H. Suoranta and A. Sivula, Nuclear magnetic resonance imaging in reversible cerebral ischaemia. Lancet 1, 294-295 (1983). 13. D. G. Penney, Acute carbon monoxide poisoning: animal models. Toxicology 62, 123-160 (1990). 14. D. G. Penney, Modifying role of plasma glucose in acute carbon monoxide poisoning. Arch. Toxicol. Suppl. 14, 239-244 (1991). 15. S. Levine, Anoxic ischemic encephalopathy in rats. Am. J. Pathol. 36, 1-17 (1960). 16. M. Ogawa, S. Shimazaki, N. Tanaka, T. Ukai and T. Sugimoto, Blood volume changes in experimental carbon monoxide poisoning. lnd. Health 12, 121-125 (1974).
17. C.-H. Nordstrom, S. Rehnchrona and B. K. Siesjo, Restitution of cerebral energy state after complete and incomplete ischemia of 30 min duration. Acta Physiol. Scand. 97, 270-272 (1976). 18. J. A. Sokal, Lack of the correlation between biochemical effects on rats and blood carboxyhemoglobin concentrations in various conditions of single acute exposure to carbon monoxide. Arch. Toxicol. 34, 331-336 (1975). 19. J. J. McGrath, K. C. Chen and P. S. Lee, Effects of intraperitoneal injection of carbon monoxide on circulating blood glucose and lactate. Fed. Proc. 38, 1313 (1979). 20. S. J. Moore, I. K. Ho and A. S. Hume, Severe hypoxia produced by concomitant intoxication with sublethal doses of carbon monoxide and cyanide. Toxicol. Appl. Pharmacol. 109, 412-420 11991). 21. V. MacMillan, Regional cerebral energy metabolism in acute carbon monoxide intoxication. Can. J. Physiol. Pharmacol. 55, 111-116 (1977). 22. S. Rehncrona, I. Rosen and B. K. Siesjo, Excessive cellular acidosis: an important mechanism of neuronal damage in the brain. Acta Physiol. Scand. 110, 435-437 (1980). 23. M. Chopp, K. M. A. Welch, C. D. Tidwell and J. A. Helpern, Global cerebral ischemia and intracellular pH during hyperglycemia and hypoglycemia in cats. Stroke 19, 1383-1387 (1988). 24. S. Lotito, P. Blondet, A. Francois, M. von Kienlin, C. Remy, J. P. Albrand, M. Oecorps, and A. L. Benabid, Correlation between intracellular pH and lactate levels in the rat brain during potassium cyanide induced metabolism blockade: a combined 31P-1Hin vivo nuclear magnetic spectroscopy study. Neurosci. Lett. 97, 91-96 (1989). 25. J. H. Garcia and K. A. Conger, Ischemic brain injuries: structural and biochemical effects, In Brain Failure and Resuscitation, ed. by A. K. Grenville and P. Safar, pp.35-54, Churchill Livingstone, London (1981). 26. B. K. Siesjo and B. Ljunggren, Cerebral energy reserves after prolonged hypoxia and ischemia. Arch. Neurol. 29, 400-407 (1973). 27. J. H. Garcia, A. S. Lossinsky, F. C. Kauffman, K. A. Conger and H. Mena, Fine structure and biochemistry of brain edema in regional cerebral ischemia, In Cerebral Vascular Diseases Eleventh Conference, ed. by T. R. Price and E. Nelson, pp.165189. Raven Press, New York (1979). 28. H. M. Pappius, Evolution of edema in experimental cerebral infarction, In Cerebrovascular Diseases, ed. by T. R. Price and E. Nelson, pp.131-139. Raven Press, New York (1979). 29. K. Muguruma, H. Nakamuta, M. Tsuji, K. Yanagita, M. Koida, H. Maeda, K. Manno, I. Narama, Y. Ogawa and Y. Hiramatsu, Biochemical and histopathological studies on a mouse brain ischemic model induced by bilateral carotid artery occlusion: Comparison with the carbon monoxide inhalation method and other ischemic or anoxic models. Nippon Yakurigaku Zasshi- Folia Pharmacologica Japonica 98, 23-29 (1991). 30. B. Sutariya, D. G. Penney, J. Barnes and C. C. Helfman, Hypothermia protects brain function in acute carbon monoxide poisoning. Vet. Hum. Toxicol. 31, 436-441 (1989).
Magnetic Resonance Imaging of the Rat Brain Following Acute Carbon Monoxide Poisoning Vishram Jalukar,' David G. Penney,'*t Michael Crowley2 and Nicholas Simpson2 'Department of Physiology, Wayne State University School of Medicine, Detroit, MI 48201, USA 2Vaitkevicius Magnetic Resonance Center, Harper Hospital, Detroit, MI 48201, USA
Key words: blood pressure; brain; carbon monoxide; cerebral; edema; glucose; hematocrit; hypotension; hypothermia; lactate; magnetic resonance imaging.
Magnetic resonance (MR) may be used for repeatedly and non-invasively imaging the brain. Until now, no studies have used this approach to study the effects of carbon monoxide (CO) poisoning in a defined animal model. Conscious, Levine-prepared female rats (unilateral carotid artery and jugular vein occlusion) were exposed to 2400 ppm CO for 90 min, with or without the infusion of 50% glucose solution; CO-stimulated increases in blood glucose and lactate occurred in both groups, while blood pressure and body temperature fell. One to four hours following termination of CO exposure, increased cortical pixel intensity, cortical surface area and brain midline shift were observed on the operated side of the brain in some rats of both groups (i.e. responders = R), providing evidence of edema. At sacrifice, 5 h following termination of CO exposure, gross water content was increased on the left side in the corresponding cortical slices in R rats, providing another measure of edema. Significant positive correlations were found between left to right pixel intensity difference and water content difference, and between the extent of midline shift and water content difference. The elevations of blood glucose and lactate concentrations, and the magnitudes of CO-induced hypothermia and hypotension were similar to those in past studies, but appeared to exert no effect on the severity of cortical edema in terms of differences in pixel intensity, surface area, midline shift or gross tissue water content. Thus, the observed differences between the R rats and the non-R rats is not explained by the available data. The results of this study demonstrate that MR imaging can detect changes in the cerebral cortices of rats given an acute toxic challenge with CO. Moreover, in responders the edema develops quickly, reaching nearly full development within 1 h after CO exposure.
INTRODUCTION
Carbon monoxide (CO) is a poisonous gas that causes death, as well as a wide variety of immediate and delayed morbid effects, with nervous system damage being the most common. In the rat, acute CO poisoning is characterized by hypotension, bradycardia, hypothermia, hemoconcentration and hyperglycemia. 1-3 A recent study using the Levine-prepared rat demonstrated increased neurological deficit and general morbidity following 90 min of C O exposure in subjects experiencing elevated blood g l u c o ~ eSimilar .~ findings were obtained from a retrospective study of human CO poisonings.s Thus, glucose level appears to be an important determinant of cerebral viability and overall survival during the hypoxia and the relative ischemia of C O poisoning. Whether elevated blood glucose permits an increased rate of glycolysis, generating more lactate and hence an increased blood and tissue acidosis, is unknown. Magnetic resonance (MR) is a recent, non-invasive technique for assessing brain structure, chemistry and the damaging effects of hypoxia, ischemia and toxic agents. Through the use of spectroscopic applications of nuclear MR, brain levels of high-energy phosphates
t Author
to whom correspondence should be addressed.
02~37W9U060407~8$09.00
01992 by John Wiley & Sons, Ltd.
have been assessed in the rat and other animal Recently, MR imaging techniques were used to study structures as small as the rat heart, following chronic C O exposure.8 While MR has been used to assess brain damage following human C O poisonings,*-" it has not been applied previously to study experimentally the induction of central nervous system damage in an animal model of CO poisoning. Some reports suggest that the technique may be superior to computerized tomography (CT), by the earlier detection of cerebral ischemic changes. l2 This study was designed to use MR imaging for non-invasively and repeatedly providing detailed information on the cerebral cortices of acutely COpoisoned rats a short period of time following termination of CO exposure. The study was carried out in an animal model in which the metabolic, cardiovascular and neurological responses to acute severe C O poisoning have been characterized previously. A major goal was to investigate the reputed deleterious role of elevated blood glucose resulting from CO exposure. A second goal was to determine whether changes in blood lactate, body temperature, blood pressure and hematocrit are associated with development of brain edema, as imaged with MR techniques. A third goal was to determine whether the magnitude of cortical edema assessed by MR methodology correlates with measurements of brain gross water content. 143'3,14
Received 12 December I991 Accepted (revised) 17 March 1992
408
V . JALUKAR ET A L .
MATERIALS AND METHODS
Table 1. Characteristics and settings of the magnetic resonance imaging instruments
Animal acquisition and treatment Female Sprague-Dawley rats, 90-120 days of age, were used. Two days prior to C O treatment, a modified Levine p r e p a r a t i ~ n ' . 'was ~ produced. The rats were anesthetized by intramuscular injection of ketamine (0.8 ml kg-')-Rompun (0.24 ml kg-'). The left common carotid artery and jugular vein were catheterized with PE-50 polyethylene tubing, placing the left cerebral hemisphere at greater ischemic risk. Both catheters were threaded subcutaneously to the nape of the neck, tied in place and plugged with Amphenol pins. Catheters were flushed twice daily with 0.1 ml of a heparin-saline solution (50 U heparin ml-*) to maintain patency. The rats were allowed to recover under close observation: they were not behaviorally different from unoperated controls. Rats ( n = 6) were exposed to 2400 ppm C O for 90 min in the conscious state. They were loosely confined in a transparent acrylic restrainer, inside a large transparent plastic bag located in a fume hood. The CO concentration was determined continuously by a Beckman model 870 infrared C O analyzer. Rectal temperature was monitored by a YSI Tele-thermometer model 43TD, using YSI 400 series probes. Heart rate and blood pressure were monitored at 0,45 and 90 min of CO exposure by a Gould model 2400 ink pen recorder, with Statham P23id pressure transducers connected to the carotid catheter. Small blood samples were withdrawn from the carotid catheter at the above times for measurements of glucose, lactate and hematocrit. A YSI 2300 STAT glucosek-lactate analyzer was used to determine blood glucose and lactate concentrations. The hematocrit was determined in duplicate by the microhematocrit method. Hyperglycemia was induced in rats ( n = 6) by the infusion of exogenous glucose (50% w/v in 0.9% saline) at a dose of 6.0 ml kg-' body weight, as described earlier.' The glucose was administered through the jugular catheter using a Harvard Apparatus Co. infusion/withdrawal pump. Infusion rates were adjusted to deliver 67% of the glucose solution in the 15 min preceding CO exposure, and the remaining 33% during CO exposure. Control rats (n = 3) were modified surgically but not treated with glucose and/or CO. Magnetic resonance imaging Immediately following CO exposure, the animals were lightly anesthetized with ketamine (0.5 ml kg-')Rompun (0.16 ml kg-') and taken to the Magnetic Resonance Center of Harper Hospital. A few unoperated control rats were also imaged under light anesthesia. The study was initiated using the 1.5 Tesla Siemens GBSII MR imaging instrument. Owing to hardware and software failures requiring protracted repairs, it was decided to complete the study using the 4.7 Tesla Bruker Biospec MR imaging unit. Rat brains were imaged in both instruments using Alderman-Grant slotted resonator coils. The instruments were autoadjusted for frequency, receiver gains and transmission power (transmission power was atten-
Parameters
Siemens GBSll 1.5
Bruker Biospec 4.7
Imaging parameters Field strength ( T ) Coil diameter (cm) Coil length (cm) Tuned (MHz) Field of view (cm) In-plane resolution (mm) Slice thickness (mm)
1.5 10.2 10.2 63 8.9 0.349 3.0
4.7 8.2 17.3 200 9.0 0.352 3.0
Spin echo imaging TE1 (ms) TE2 (ms) Slice distance (mm) Slice gap (mm) Acquisitions Imaging time (min)
22 or 26 90 0.05 0.15 2 21
26 90 1.75 5.25 1 17
uated 25 dB subsequent to autoadjust in the Bruker instrument) (see Table 1). Scout images were used to set the orientation of the slices in the spin echo sequence in order to obtain true coronal images, as shown below. Spin echo images were acquired 1, 2, 3 and 4 h after removal from CO. Three adjacent images corresponding to rostral, middle and caudal sections of the brain were analyzed for midline shift, total cortex area and relative edema index, using standard Siemens or Bruker software. Midline shift was calculated by measuring the diameter of the total brain and of each hemisphere. The relative edema index was calculated by measuring the average pixel intensity within a circle of radius 0.210 cm from corresponding portions of each cortical hemisphere. Total cortex area was defined by drawing a freehand outline of each cortical hemisphere. Rats were sacrificed after imaging by administering an overdose of ketamine through the jugular catheter. The cortex was immediately removed by dissection and sectioned into three slices per hemisphere. The sections were dried at 60°C for 48 h. Wet and dry weights were measured using a Mettler H35AR analytical balance. Data analysis was carried out on Macintosh microcomputers using Excel and Cricket-Graph programs. Most values are means f SEM. Student's #-test was used for statistical analysis, with P values =s 0.05 considered to be significant. RESULTS Upon 'H imaging, some CO-exposed rats of both groups (CO alone and CO+glucose) showed marked edema and swelling of the left cerebral hemisphere ('responders'), while others did not ('non-responders'). The presence of cerebral edema is indicated by increasing brightness in the region of fluid accumulation and of a rightward shift of the brain midline (Fig. 1C,
MR IMAGING IN CO POISONING
E). This is illustrated by the plots of pixel intensity superimposed upon each cortical cross-section (Fig. l A , B, D, E). Control rats, non-responders and unoperated animals (not shown) showed no right versus left brightness or midline differences (Fig. l A , B, D). The photographs shown were judged to be representative of the responses by controls, responders and nonresponders. Note that, by convention, the left side of the brain is on the right side in each photograph. The data presented are a fraction of the total number of rats treated and imaged. Some rats were not imaged satisfactorily and thus were not included in the final data set. No selection criteria were applied other than completeness. Evidence of cortical edema is also seen in Fig. 2 for individual animals, in terms of intensity (A), midline shift (B), differences in area (C) and in water content (D), which was determined afterwards on corresponding brain slices. The values for intensity, midline shift and area at 1 and 4 h of recovery, and for water content ( 5 h only). are seen in Table 2. There was no clear time dependence with regard to the development of cerebral edema in responders, either in terms of the differences in intensity or in midline position (Fig. 3A, B). In most cases, increased intensity and the midline shift were already present in responders by 1 h of recovery. Correlations of both intensity difference and the midline shift at 1 h of recovery with the water content difference of the corresponding brain slices were each statistically significant (Fig. 4A, B). The correlation of midline shift at 4 h with water content difference was also significant. When considered in terms of mean values, the responders tended to show area and intensity difference values greater than the nonresponders (but not significant) (Fig. 5). The water content difference of the responders was significantly elevated above the non-responders (Fig. 6). There was no difference in water content between the nonresponders and a large group of unoperated controls. The metabolic and physiological parameters of the experimental subjects are shown in Table 3. They are displayed in two ways: as rats that received CO alone and CO+glucose, and as responders and nonresponders. Glucose infusion raised blood glucose to 241 mg d1-I 45 min after beginning CO exposure and to 168 mg dl-' in rats receiving CO alone. Blood lactate in the rats receiving glucose was raised to 170 mg dl-' 45 min after beginning C O exposure and to 151 mg dl-' in rats receiving CO alone. Average blodd glucose and lactate values of all seven animals prior to CO exposure were 97 and 9 mg dl-', respectively. Also at this time, mean arterial blood pressure was 110 mmHg, body temperature was 37.6"C. heart rate was 470 beats min-' and the hematocrit was 41.7%. The responder group consisted of two CO and two CO+glucose rats, while the non-responders consisted of one CO and two CO+glucose rats. There were no significant differences ( P > 0.05) between the responders and the non-responders with respect to blood glucose or lactate concentrations, or change in mean arterial blood pressure, body temperature. heart rate or hematocrit (Table 3). None the less, lactate concentration and the fall in blood pressure and body temperature tended to be greater in the responders.
409
Neither blood glucose nor lactate concentration or hypotension during CO exposure appeared to be correlated with the severity of cortical edema. ~~
~~
DISCUSSION
The results of this study demonstrate that MR imaging can detect changes in the cerebral cortices of Levineprepared rats acutely poisoned with CO. The edema develops quickly, being well developed within 1 h after exposure ends. It develops in this model with or without the addition of exogenous glucose, and is apparently not related to the magnitude of hyperglycemia, lactic acidemia, hypothermia, hypotension or hemoconcentration. Thus, the observed differences between the responders and the non-responders is not explained by the available data. On the other hand, even if correlations had been established, it would not be possible to state with certainty that the changes in these parameters were responsible for the cortical edema, or that they resulted from it. It is possible that the non-responders had adequate bilateral cortical perfusion despite the surgical modification, thus obviating the development of unilateral cerebral hypoxia sufficient to result in edema. This may be attributable to the anatomical variations among rats. Explanations involving differences in oxygen delivery, metabolism, etc. are also possible. Significant associations of edema development with cardiovascular and other physiological parameters may resolve if a larger number of animals were to be studied. Hypothermia, hypotension, bradycardia, hemoconcentration and altered blood glucose concentration are the usual effects of acute severe CO poisoning in the The loss of normal body temperature results from decreased heat production through an inhibition of oxidative metabolism and an increased heat loss due to peripheral vasodilation. l 3 Decreased arterial blood pressure is a function of the peripheral vasodilation, and depressed pump function is due to both decreased heart rate and stroke volume. The rise in the hematocrit is presumed to be due to the loss of plasma volume resulting from an increased endothelial permeability ."j Increased hematocrit and lowered body temperature act to increase blood viscosity. This, along with compromised cardiac function, contributes to inadequate cerebral perfusion. In the Levine-prepared rat, the oligemia is more extreme during CO exposure on the operated side of the brain, leading to edema and behavioral evidence of unilateral brain dysfunction. I Previous studies in this laboratory showed that elevated blood glucose is associated with poor outcome in the CO-poisoned rat."' This was the case whether the elevated glucose resulted from CO exposure' or from the infusion of exogenous glucose.3 The blood glucose concentrations achieved during CO exposure in the present study in the C O alone and CO+glucose rats were of a magnitude similar to earlier studies. None the less, it should be noted that the responders showed no higher blood glucose concentration than the nonresponders. Elevated blood glucose is also reported to be deleterious in human acute CO p o i s ~ n i n g . ~ The mechanism by which hyperglycemia produces
410
V . JALUKAR ET A L
Figure 1. TZ-weighted proton magnetic resonance echo 1 images of rat cerebral cortices: (A) Control; (6) CO-exposed non-responder; (C) CO-exposed responder; (D) CO-exposed + glucose-treated non-responder; (E) CO-exposed + glucose-treated responder. All but control are 4 h post-CO exposure. The left side of the brain is on the right side of the picture. All but image C were acquired on the Bruker 4.7T instrument: image C was acquired on the Siemens 1.5T instrument. Pixel intensity is plotted in all but panel C.
41 1
MR IMAGING IN CO POISONING
I
100, 40
A.
30 20
80 A
0
a
-
s - .
10
m
730.1 807.2
0 812.1
60-
.-c
-10
40-
a
.
s
20-
w
a
I
4
-20
0
2
1
3
5
4
8
B. 61
I
-204
I
a o -4
1
0
co
Control
3
2
5
4
Recovery Time (hrs.)
CO + Glucose
Figure 3. Differences between left and right cerebral hemispheres of rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose, at 1,2,3 and 4 h post-exposure: (A) mean pixel intensity; (6)midline position. Responders are denoted by filled symbols and non-responders by open symbols. Numbers are animal identifiers.
Figure 2. Differences between left and right cerebral hemispheres of non-exposed control rats and of rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose: (A) mean pixel intensity; (6)midline position (C) total cortical area; (D)water content. (A-C) 1 h postC0 exposure; (D)ca. 5 h post-CO exposure.
Table 2. Differences between left and right cerebral hemispheres of control rats and rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose twice during recovery Control rats
CO rats
CO+glucose rats
Parametermime
702
A
721
724
730-1
807-1
807-2
812-1
-2.11
4.12
12.79 19.78
30.17
-10.09 5.36
41.38 86.36
2.13 7.61
10.17 37
1.4 9.33
1h 4h
0
0.68
3.95 5.32
1.52 3.44
0.71 0.34
3.62 5.47
1.57 1.27
2.53 3.95
-2.52 0
Area
I h
6.36
-16.12
(YO)
4h
35.49 40.96
5.2 14.86
0 -3.66
39.63 55.1
15.84 -9.46
12.09 13.24
-3.74 7.25
0.06
0
4.22
3.20
-0.49
4.45
0.43
3.07
0.31
A Midline (YO)
A
708
1h 4h
Intensity
l%)
A
821
Water contenta (YO)
Water content was determined once at ca. 5 h of recovery.
brain damage has been tied to the accumulation of lactate and to the resulting acidosis. l4 Acidosis could be more severe in CO poisoning than in pure ischemic situations, since some blood flow continues to supply glucose and to generate lactate, although the flowing blood also continues to wash the lactate away. In other models, incomplete ischemia has been shown to have a worse outcome than complete ischemia.17 In any event, blood lactate concentration was observed to increase some 16-fold during CO exposure in the present study,
from 9 mg dl-1 (pre-CO) to ca. 150 mg dl-* (during CO). This must have been accompanied by a sharp fall in blood pH; unfortunately, we did not measure pH. SokalIs found that blood lactate increased from 9 to 62 mg dl-' in rats exposed to 4000 ppm CO for 40 min, or an 18-fold increase. Considering that the CO, concentration that he used was nearly double that of the present study, the similarity in response was excellent. Blood pH fell to 6.85. Similar changes in blood lactate were reported by McGrath et ~ 1 . in ' ~rats
412
V. JALUKAR ET A L . I
1001
80
C 0,
1 .
0
5.0
..
r = 0.86. WO.01
-20 -0.01
0.00
0.01
0.02
0.03
0.04
s i -1.01
0.05
Responders
Nonresponders
Controls
Figure6. Cortical water content of rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose, and non-exposed controls. Exposed rats have been grouped as responders and non-responders. The numbers of rats in each group, and the means t SE are given. Comparing responders to non-responders and controls; ** P < 0.01.
-001
000
0 01
0 02
0 03
0 04
0 05
A Water Content (%I
Figure4. Regression plots using data from left and right cerebral hemispheres of rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose, and of nonexposed controls: (A) difference in mean pixel intensity (lef’t minus right) at 1 (open symbols) and 4 h (closed symbols) post-CO exposure vs. the difference in cortical water content (left minus right); (B)differnece in midline position vs. the difference in cortical water content (left minus right). Intensity and midline position data obtained at 1 (open symbols) and 4 h (closed symbols) post-CO exposure; water content obtained at ca. 5 h post-CO exposure. Midline position was calculated by subtracting the radius of the left hemisphere from the radius of the brain.
80 70
T
I
-30
Responders
Nonresponders
Controls
Figure5. Cortical area and mean pixel intensity of the cortex in rats exposed to 2400 ppm carbon monoxide, with and without exogenous glucose, and non-exposed controls, at 1 and 4 h post-exposure. Values for controls determined once. The CO-exposed rats are grouped as responders and nonresponders. The numbers of rats in each group are given. Vertical bars indicate standard error.
acutely inhaling CO. Using mice inhaling 3500 pprn CO, Moore et al.”’ found that blood lactate reached 90 mg dl-’ after 11.5 min while pH fell to only 7.25, the more modest acidosis probably being due to the short duration of exposure. Measurements of regional brain lactate concentration have shown it to increase to similar high levels in anesthetized rats ventilated with up to 20000 ppm CO.” In the past, a number of investigators have suggested that hypoxic/ischemic brain damage is the direct result of excessive lactate production and the attendant acidosis,22 and that brain lactate can be used to calculate brain pH. Recent MR studies of brain lactate and pH, however, suggest that the two become dissociated during hypoxia/ischemia.23.24Although there were no differences in the blood lactate levels in our rats, it is still possible that brain pH was different in the responders versus the non-responders. Nevertheless, recent studies in our laboratory have found no difference in peak blood lactate concentration between rats that died during and following CO exposure and those that survived, whether or not exogenous glucose was given (Sutariya L? Penney, unpublished data). Administration of 50% glucose solution, which raised the blood glucose level to 325-450 mg dl-’ during CO exposure, failed to elevate further the peak blood lactate level. This finding corroborates the present study, where the administration of glucose failed to elevate significantly the blood lactate level and where responders showed no higher lactate level than non-responders. Of course, it is possible that tissue lactate concentration in the cortical region that sustained edema departed significantly from that of the carotid blood. We found an increase in cortical water content of as much as 4.5% on the operated side of the brain in responder rats. This value agrees well with an earlier study with Levine-prepared rats exposed to 2700 ppm CO, in which the difference in cerebral water content (left minus right) was significantly correlated with behaviorally derived neurological scores. * The impaired rats showed obvious evidence of unilateral brain dysfunction, e.g. one-sided weakness and circling. Neurological assessment in a similar fashion was not possible in the present study because of the necessity to anesthetize the rats during
413
M R IMAGING IN CO POISONING
Table 3. Metabolic and physiological parameters of rats exposed to 2400 ppm carbon monoxide alone and 2400 ppm carbon monoxide + glucose, of the rats of both groups that showed evidence of cortical edema (responders) and of rats not showing cortical edema (non-responders)." Parameterrrime (min)
CO (3)
CO+Glucose (4)
Responders (4Ib
* 14.0
Non-responders (3)c
Glucose (mg dl '1 Lactate (mg dl-')
+45
167.7
33.2
241 t 50.6
180.8
+45 +90
151.0 2 12.7 137.0 ? 20.0
69.8 t 19.0 50.5 ? 13.3
158.3 15.7 154.0 5 13.1
66.3 ? 21.7 32.3 5 18.0
A
+90
-36.0 5 1.0
45.4 5 4.5
-43.5
4.4
39.8
9.5
- 1 4 t 33.8
Av. blood
pressure (mmHg) A Heart rate (beats min-'1
A A
+90
-43
?
?
17
Body temp. ("C)
+90
-4.2 t 0.3
Hematocrit
+90
5.1 f 0.5
2.3 2 99.3 -3.4 5 0.4 6.0
?
-19.5
-4.2 t 0.2
* 0.6
5.2
t_
0.3
248
-3.0
?
?
?
78.0
7.5
0.4
6.2 5 0.9
(YO)
a
Values are means t SEM. Responders: two CO and two CO+glucose. Non-responders: one CO and two CO+glucose.
MR imaging and immediately prior to cortical tissue removal for the measurement of water content. The finding in normal rats of constant cortical water content, with very little variability between the two hemispheres, suggests little influence of potentially confounding factors (e.g. estrous cycle). The brain is relatively resistant to injury due to hypoxia, but edema develops almost immediately in response to ischemia." Hypoxia is almost always accompanied by a decrease in blood pressure. In fact, severe hypoxia leads to cardiovascular failure resulting in global ischemia, thus the effects of hypoxia alone are difficult to assess.'5 In cases where blood pressure is maintained, there is no evidence of brain abnormalities in experimental animals subjected to hypoxia.2h Brain edema results from the retention of water.27 Studies on ischemia suggest that there is a threshold phenomenon for edema formation.28One might expect to find a correlation between the CO-induced hypotension observed and edema formation. The absence of such a correlation in our rats indicates that there may be variability between individuals in the collateral circulation. The existence of similar collateral circulatory variability was inferred in mice exposed to C O or ether.2' Further study of cerebral blood flow during CO exposure is indicated. We have shown that hypothermia. hypotension and hemoconcentration are usual events during CO exposure in the rat,' and indeed that hypothermia confers protection in terms of survival and neurological outcome."' Inexplicably, in the present study, body temperature fell by more in the responders than in the non-responders, although the difference was of marginal statistical significance. Two human cases of CO-induced brain damage reportedly have been diagnosed with MR imaging, where lesions were found in the globus pallidus: in one case, bilateral cortical damage also was found." In a second case report, damage to the globus pallidus caused by
CO exposure was determined using both MR and CT imaging techniques.'" More recently, Tuchman et al. using MR imaging, found bilateral increased signal intensity in the anterior thalamus of a child poisoned with CO. An earlier CT scan had revealed no abnormality, suggesting the advantage of MR imaging. Other studies" suggest that MR imaging may reveal cerebral ischemic changes earlier than other techniques. In summary, this study of acute severe C O poisoning in the Levine-prepared rat resulted in the following findings: (i)
development of edema as evidenced by increased pixel signal intensity, surface area and midline bulging of the cerebral cortex on the operated side of the brain, 1-4 h following the termination of CO exposure in responder rats; (ii) development of edema as evidenced by increased gross water content in the corresponding cortical slices in responder rats, approximately 5 h following the termination of CO exposure; (iii) demonstration of a significant positive correlation between cortical pixel signal intensity difference as a measure of edema and water content difference and between midline shift as a measure of edema and water content difference; (iv) within the limitations posed by the small number of animals used, the increase in blood glucose and lactate showed no correlation with the severity of cortical edema, as judged by differences in pixel signal intensity, surface area, midline position or gross water content.
Acknowledgements This work was supported by an American Heart Association of Michigan Summer Research Fellowship to Mr Vishram Jalukar. We wish to thank the Magnetic Resonance Center, Harper Hospital, for providing the instrument time to carry out this study. We also wish to thank William Negendank, M D and Jeffrey Evelhoch, PhD.
414
V. JALUKAR ET A L .
REFERENCES
1. D. G. Penney, K. Verma and J. A. Hull, Cardiovascular, metabolic and neurologic effects of acute carbon monoxide poisoning in the rat. Toxicol. Lett. 45, 207-213 (1989). 2. D. G. Penney, P. Sharma, 8. B. Sutariya, and 6. G. Nallamothu, Development of hypoglycemia is associated with death during carbon monoxide poisoning. J. Crit. Care 5, 1-7 (1990). 3. D. G. Penney, C. C. Helfman, J. C. Dunbar and L. E. McCoy, Acute severe carbon monoxide poisoning in the rat: effects of hyperglycemia and hypoglycemia on mortality, recovery and neurologic deficit. Can. J. Physiol. Pharmacol. 69, 1168-1177 (1991). 4. D. G. Penney, C. C. Helfman, J. A. Hull, J. C. Dunbar and K. Verma, Elevated blood glucose is associated with poor outcome in the carbon monoxide-poisoned rat. Toxicol. Lett 54, 287-298 (1990). 5. D. G. Penney, Hyperglycemia exacerbates brain damage in acute severe carbon monoxide poisoning. Med. Hypoths. 27, 241-244 (1988). 6. W. Auffermann, N. Derugin, S. Wu, W. W. Parmley, C. B. Higgins and J. Wikman-Coffelt, Effect of reduced perfusion pressure in cardiac energetics and function in pressure overload hypertrophy -3'P-NMR in rats. Circulation, Suppl. 76 (IV), IV-439 (1987). 7. K. L. Behar, J. A. den Hollander, 0. A. C. Petroff, H. P. Hetherington, J. W. Prichard and R. G. Shulman, Effect of hypoglycemic encephalopathy upon amino acids, highenergy phosphates, and pHi in the rat brain in vivo: detection by sequential 'H and 31P NMR spectroscopy. J. Neurochem. 44, 1045-1 055 (1985). 8. G. A. Bambach, D. G. Penney and N. G. Negendank, In situ assessment of the rat heart during chronic carbon monoxide exposure using magnetic resonance imaging. J. Appl. Toxicol. 11, 43-49 (1991). 9. A. L. Horowitz, R. Kaplan and G. Sarpel, Carbon monoxide toxicity: MR imaging in the brain. Radiology 162, 787-788 (1987). 10. G. Meucci, G. Rossi and M. Mazzoni, A case of transient choreoathetosis with amnesic syndrome after acute carbon monoxide poisoning. ltal. J. Neurol. Sci. 10, 513-517 (1989). 11. R. F. Tuchman, F. G. Moser and S. L. Moshe, Carbon monoxide poisoning: bilateral lesions in the thalamus on MR imaging of the brain. Pediatr. Radio/.20,478-479 (1990). 12. J. T. Sipponen, M. Kaste, R. E. Sepponen, T. Kuurne, H. Suoranta and A. Sivula, Nuclear magnetic resonance imaging in reversible cerebral ischaemia. Lancet 1, 294-295 (1983). 13. D. G. Penney, Acute carbon monoxide poisoning: animal models. Toxicology 62, 123-160 (1990). 14. D. G. Penney, Modifying role of plasma glucose in acute carbon monoxide poisoning. Arch. Toxicol. Suppl. 14, 239-244 (1991). 15. S. Levine, Anoxic ischemic encephalopathy in rats. Am. J. Pathol. 36, 1-17 (1960). 16. M. Ogawa, S. Shimazaki, N. Tanaka, T. Ukai and T. Sugimoto, Blood volume changes in experimental carbon monoxide poisoning. lnd. Health 12, 121-125 (1974).
17. C.-H. Nordstrom, S. Rehnchrona and B. K. Siesjo, Restitution of cerebral energy state after complete and incomplete ischemia of 30 min duration. Acta Physiol. Scand. 97, 270-272 (1976). 18. J. A. Sokal, Lack of the correlation between biochemical effects on rats and blood carboxyhemoglobin concentrations in various conditions of single acute exposure to carbon monoxide. Arch. Toxicol. 34, 331-336 (1975). 19. J. J. McGrath, K. C. Chen and P. S. Lee, Effects of intraperitoneal injection of carbon monoxide on circulating blood glucose and lactate. Fed. Proc. 38, 1313 (1979). 20. S. J. Moore, I. K. Ho and A. S. Hume, Severe hypoxia produced by concomitant intoxication with sublethal doses of carbon monoxide and cyanide. Toxicol. Appl. Pharmacol. 109, 412-420 11991). 21. V. MacMillan, Regional cerebral energy metabolism in acute carbon monoxide intoxication. Can. J. Physiol. Pharmacol. 55, 111-116 (1977). 22. S. Rehncrona, I. Rosen and B. K. Siesjo, Excessive cellular acidosis: an important mechanism of neuronal damage in the brain. Acta Physiol. Scand. 110, 435-437 (1980). 23. M. Chopp, K. M. A. Welch, C. D. Tidwell and J. A. Helpern, Global cerebral ischemia and intracellular pH during hyperglycemia and hypoglycemia in cats. Stroke 19, 1383-1387 (1988). 24. S. Lotito, P. Blondet, A. Francois, M. von Kienlin, C. Remy, J. P. Albrand, M. Oecorps, and A. L. Benabid, Correlation between intracellular pH and lactate levels in the rat brain during potassium cyanide induced metabolism blockade: a combined 31P-1Hin vivo nuclear magnetic spectroscopy study. Neurosci. Lett. 97, 91-96 (1989). 25. J. H. Garcia and K. A. Conger, Ischemic brain injuries: structural and biochemical effects, In Brain Failure and Resuscitation, ed. by A. K. Grenville and P. Safar, pp.35-54, Churchill Livingstone, London (1981). 26. B. K. Siesjo and B. Ljunggren, Cerebral energy reserves after prolonged hypoxia and ischemia. Arch. Neurol. 29, 400-407 (1973). 27. J. H. Garcia, A. S. Lossinsky, F. C. Kauffman, K. A. Conger and H. Mena, Fine structure and biochemistry of brain edema in regional cerebral ischemia, In Cerebral Vascular Diseases Eleventh Conference, ed. by T. R. Price and E. Nelson, pp.165189. Raven Press, New York (1979). 28. H. M. Pappius, Evolution of edema in experimental cerebral infarction, In Cerebrovascular Diseases, ed. by T. R. Price and E. Nelson, pp.131-139. Raven Press, New York (1979). 29. K. Muguruma, H. Nakamuta, M. Tsuji, K. Yanagita, M. Koida, H. Maeda, K. Manno, I. Narama, Y. Ogawa and Y. Hiramatsu, Biochemical and histopathological studies on a mouse brain ischemic model induced by bilateral carotid artery occlusion: Comparison with the carbon monoxide inhalation method and other ischemic or anoxic models. Nippon Yakurigaku Zasshi- Folia Pharmacologica Japonica 98, 23-29 (1991). 30. B. Sutariya, D. G. Penney, J. Barnes and C. C. Helfman, Hypothermia protects brain function in acute carbon monoxide poisoning. Vet. Hum. Toxicol. 31, 436-441 (1989).
